Electron microscopic immunohistochemical features of the nuclear inclusions in motor neurons and scrotal skin epidermal cells of SBMA patients. Dense granular aggregation of AR-immunoreactive material without a limiting membrane was found in both motor neurons ( A to C ) and scrotal skin epidermal cells ( D to F ). Nucleoli, which are larger than the inclusions, are clearly distinguished. The same inclusions observed light microscopically ( A and D ) are demonstrated electron microscopically ( B , C , E , and F ). AR(N-20) was used as the first Ab. Magnification: A , 720 ϫ ; D , 1,000 ϫ ; B and E , 2,000 ϫ ; C , 10,000 ϫ ; F , 15,000 ϫ . 

Contexts in source publication

Context 1

... whom were processed for frozen samples and all five fixed in formalin). These patients were 54 to 82 years of age at death, and each showed a typical clinical phenotype of SBMA, with dysphagia, bulbar and extremity muscle weakness, and atrophy with fasciculation. Gynecomastia and diabetes mellitus were present in four patients. Du- ration from onset to death was 9 to 23 years, and the causes of death were empyema, bronchiectasis, and gastric cancer. The CAG repeat lengths of the AR gene determined in blood samples were 40 to 52. Tissue samples for immunohistochemical analysis were obtained at autopsy, frozen in liquid nitrogen, and stored at Ϫ 80°C or were fixed in 10% buffered formalin and processed for paraffin section. The pathological features of these cases were also typical for SBMA, 2,3 with minimal variation in extent among the patients; the spinal, bulbar and pontine motor neurons were extensively depleted, with mild glio- sis; sensory neurons were mildly affected, with occa- sional Nageotte’s nodules; the posterior column of the spinal cord was depleted in a rostrally accentuated manner; the muscles were chronically denervated, and the sural nerve myelinated fibers were moderately depleted. Testicular atrophy and fatty liver changes were also present. Other portions of the central nervous system and visceral organs were normal except for pulmonary infec- tion in all patients and gastric cancer in one patient. Control tissue samples were obtained from four male autopsied patients ages 54 to 71 years, who died of nonneurological diseases. The AR CAG repeat lengths of the controls were 19 to 24. All autopsies were performed within 6 hours post- mortem. Several polyclonal and monoclonal Abs that specifically recognize the AR protein were used in this study (Figure 1): 2F12 (mouse monoclonal Ab (immunoglobulin (Ig) G), NovoCastra, Newcastle, UK), generated against a re- combinant protein of 321 amino acids from the N terminus of the human AR; PG-21 (rabbit polyclonal Ab (IgG), Affinity BioReagents, Golden, CO) and AR(N-20)(rabbit polyclonal Ab (IgG), Santa Cruz Biotechnology, Santa Cruz, CA), which recognize 21 and 20 amino acid resi- dues of the N terminus of the AR, respectively; AR52 (rabbit polyclonal Ab (IgG), kindly provided by Dr. E. Wilson, University of North Carolina, Chapel Hill, NC), which recognizes the DNA-binding domain of the AR; and 5F4 (mouse monoclonal Ab (IgM), kindly provided by Dr. T Demura, Department of Urology, Hokkaido Univer- sity, Hokkaido, Japan) and AR(C-19) (rabbit polyclonal Ab (IgG), Santa Cruz Biotechnology), which recognizes the C terminus of the human AR. Characterization and binding specificities of all of these Abs to human AR were previously described. 33,37–39 Anti-ubiquitin Ab (rabbit polyclonal Ab (IgG), Dakopatts, Glostrup, Denmark) was also used. Cryostat sections of 8 ␮ m were prepared from the frozen tissues of SBMA patients and controls, quickly dried, and lightly fixed with Zamboni fixative for 10 minutes. Then the tissue sections were washed, blocked with normal horse serum (1:20), and incubated with Abs against AR, ubiquitin, or affinity-purified mouse IgG1 at concentrations of 0.5 to 4 ␮ g/ml. Endogenous peroxidase was blocked by preincubation of tissue sections with 0.3% H 2 O 2 in meth- anol for 30 minutes. Endogenous biotin was also blocked by incubation with an avidin-biotin blocking kit (Vector Laboratories, Burlingame, CA). Immune complexes were visualized using the avidin-biotinylated horseradish peroxidase system (Elite Vector kit from Vector Laboratories) and 3,3 Ј -diaminobenzidine (Dakopatts) substrate. Sec- tions were counterstained with methyl green. For paraffin- embedded samples, the 4- ␮ m tissue sections were deparaffinized in xylene, hydrated with alcohol, and then heated in a microwave oven for 5 minutes. The tissue sections were then processed in the same way as those for the frozen tissue sections. Immune complexes were visualized using the tyramide signal amplification system (New England Nuclear, Boston, MA) following the manu- facturer’s protocol. For electron microscopic immunohistochemistry, buffered formalin-fixed, paraffin-embedded tissue sections were immunostained with Abs against AR and then incubated with horseradish peroxidase-labeled second Ab (Amersham, Poole, UK). The tissue sections were then visualized with 3,3 Ј -diaminobenzidine (Dakopatts), fixed with 2% osmium tetroxide in 0.1 mol/L phosphate buffer, pH 7.4, and dehydrated in an alcohol gradient and embedded in epoxy resin, from which ultrathin sections were obtained and then observed under an electron micro- scope (Hitachi H-7000). CAG repeat length of the AR gene was determined on the autopsied tissue samples using methods described previously. 5,40 Nuclear inclusions were detected in the spinal motor neurons, neurons in the motor nucleus of the trigeminal nuclei in the pons, and neurons of the hypoglossal nuclei of the medulla oblongata (Figure 2 (see also Figure 4), Table 1). Nuclear inclusions were not detected in the neurons of cerebral cortex, caudate, thalamus, pallidum, cerebellar cortex, dentate nucleus nuclei other than motor nuclei in the brain stem, intermediolateral and Clarke’s nuclei of the spinal cord, or dorsal root ganglion neurons. Nuclear inclusions similar to those in the motor neurons were detected in the scrotal skin, dermal skin, kidney, heart, and testis (Figures 3 and 4, Table 1). The inclusions were not seen in the liver, spleen, or muscle. Both neural and nonneural nuclear inclusions were strongly stained by PG-21 and AR(N-20), but not by 2F12, AR52, 5F4, or AR(C-19) (Table 2). The inclusions were spherical structures of 1 to 5 ␮ m in diameter, and those in the nonneural tissues were generally smaller in size than those in the neurons. They were ubiquitinated (Figure 3, Tables 1 and 2). There were no detectable inclusions in cytoplasm of the neural or nonneural tissues. Most com- monly we saw one inclusion in the nucleus of the individ- ual cells, but two or three inclusions per cell were also observed (Figures 2 to 4) in both the neural and nonneural tissues. The frequency of the nuclear inclusions in the spinal motor neurons was 8.39 Ϯ 5.43% of total remain- ing motor neurons. The frequencies of nuclear inclusions in nonneural tissues were 0.93 Ϯ 0.50% in scrotal skin epidermal cells, 0.40% in nonscrotal skin epidermal cells, and 0.59 Ϯ 0.02% in kidney tubular cells, and nuclear inclusions were observed only occasionally in the heart and testis. Electron microscopically, these inclusions consisted of granular dense aggregates of AR-positive materials without limiting membrane (Figure 4), and the morphological appearance was quite similar among the inclusions in the spinal motor neurons, scrotal skin, and kidney. There was no evidence of filamentous structures as reported in HD 27 and MJD 30 (Figure 4). The morphological appearance of the motor neurons and nonneural cells with nuclear inclusions was indistin- guishable from those without inclusions. Neural and nonneural tissues from four control cases were also examined in the same manner as that for SBMA cases, but the inclusions were not seen in the control individuals. The present study demonstrates that nuclear inclusions of the mutant AR protein are present in the motor neurons of the spinal cord and the brain stem; moreover, similar inclusions are also detected in certain nonneural tissues of the SBMA patients, indicating that the occurrence of the nuclear inclusions is not restricted to the affected neural tissues, but is also found in certain nonneural tissues. The electron microscopic appearance of the AR-positive dense aggregates with similar granular size without limiting membrane was common to both neural and nonneural tissues. Furthermore, a characteristic selective staining pattern of the nuclear inclusions seen with only Abs recognizing N-terminal 20 and 21 amino acids of the AR protein was also common among the neural and nonneural tissues. These observations strongly suggest that same mechanism is involved in formation of AR- positive components in the nuclear inclusions in both the neural and nonneural tissues. The immunostaining pattern suggests that only a small portion of the N terminus of the AR protein is available as an epitope in the nuclear inclusions, and other portions of the AR may be masked within the inclusions of the mutant protein, or alternatively, the AR protein may be cleaved by proteolytic activity resulting in N-terminal fragments that participate in the aggregation, as was suggested in other polyglutamine diseases. 26 –36,41 In addition, these AR nuclear inclusions are ubiquitinated in the nonneural as well as neural tissues, indicating that the nuclear inclusion is a pathological structure of the mutant AR, even in the nonneural tissues, where the pathological involvement is not appar- ent. Our electron microscopic and light microscopic immunohistochemical data indicate that nuclear inclusions in both motor neurons and nonneural tissues are identical in morphological and immunochemical features. Furthermore, absence of filamentous structures in the nuclear inclusion of SBMA was different from observations in HD 27 and MJD. 30 This difference may suggest that the pathway of the nuclear aggregation is different among the different protein products or, alternatively, may rep- resent variances in sample preparation. In polyglutamine diseases that have been analyzed to date, the nuclear inclusions have been shown to occur selectively in neurons of the affected brain regions. The selective occurrence of nuclear inclusions in the affected cells of central nervous system in SBMA that we observed in this study agrees well with observations of HD, MJD, SCA1, and DRPLA, 27–32 as well as our previous observations in SBMA. However, the appearance of similar nuclear inclusions in the nonneural tissues observed in this study is novel. It is an important ...

Context 2

... Several polyclonal and monoclonal Abs that specifically recognize the AR protein were used in this study (Figure 1): 2F12 (mouse monoclonal Ab (immunoglobulin (Ig) G), NovoCastra, Newcastle, UK), generated against a re- combinant protein of 321 amino acids from the N terminus of the human AR; PG-21 (rabbit polyclonal Ab (IgG), Affinity BioReagents, Golden, CO) and AR(N-20)(rabbit polyclonal Ab (IgG), Santa Cruz Biotechnology, Santa Cruz, CA), which recognize 21 and 20 amino acid resi- dues of the N terminus of the AR, respectively; AR52 (rabbit polyclonal Ab (IgG), kindly provided by Dr. E. Wilson, University of North Carolina, Chapel Hill, NC), which recognizes the DNA-binding domain of the AR; and 5F4 (mouse monoclonal Ab (IgM), kindly provided by Dr. T Demura, Department of Urology, Hokkaido Univer- sity, Hokkaido, Japan) and AR(C-19) (rabbit polyclonal Ab (IgG), Santa Cruz Biotechnology), which recognizes the C terminus of the human AR. Characterization and binding specificities of all of these Abs to human AR were previously described. 33,37–39 Anti-ubiquitin Ab (rabbit polyclonal Ab (IgG), Dakopatts, Glostrup, Denmark) was also used. Cryostat sections of 8 ␮ m were prepared from the frozen tissues of SBMA patients and controls, quickly dried, and lightly fixed with Zamboni fixative for 10 minutes. Then the tissue sections were washed, blocked with normal horse serum (1:20), and incubated with Abs against AR, ubiquitin, or affinity-purified mouse IgG1 at concentrations of 0.5 to 4 ␮ g/ml. Endogenous peroxidase was blocked by preincubation of tissue sections with 0.3% H 2 O 2 in meth- anol for 30 minutes. Endogenous biotin was also blocked by incubation with an avidin-biotin blocking kit (Vector Laboratories, Burlingame, CA). Immune complexes were visualized using the avidin-biotinylated horseradish peroxidase system (Elite Vector kit from Vector Laboratories) and 3,3 Ј -diaminobenzidine (Dakopatts) substrate. Sec- tions were counterstained with methyl green. For paraffin- embedded samples, the 4- ␮ m tissue sections were deparaffinized in xylene, hydrated with alcohol, and then heated in a microwave oven for 5 minutes. The tissue sections were then processed in the same way as those for the frozen tissue sections. Immune complexes were visualized using the tyramide signal amplification system (New England Nuclear, Boston, MA) following the manu- facturer’s protocol. For electron microscopic immunohistochemistry, buffered formalin-fixed, paraffin-embedded tissue sections were immunostained with Abs against AR and then incubated with horseradish peroxidase-labeled second Ab (Amersham, Poole, UK). The tissue sections were then visualized with 3,3 Ј -diaminobenzidine (Dakopatts), fixed with 2% osmium tetroxide in 0.1 mol/L phosphate buffer, pH 7.4, and dehydrated in an alcohol gradient and embedded in epoxy resin, from which ultrathin sections were obtained and then observed under an electron micro- scope (Hitachi H-7000). CAG repeat length of the AR gene was determined on the autopsied tissue samples using methods described previously. 5,40 Nuclear inclusions were detected in the spinal motor neurons, neurons in the motor nucleus of the trigeminal nuclei in the pons, and neurons of the hypoglossal nuclei of the medulla oblongata (Figure 2 (see also Figure 4), Table 1). Nuclear inclusions were not detected in the neurons of cerebral cortex, caudate, thalamus, pallidum, cerebellar cortex, dentate nucleus nuclei other than motor nuclei in the brain stem, intermediolateral and Clarke’s nuclei of the spinal cord, or dorsal root ganglion neurons. Nuclear inclusions similar to those in the motor neurons were detected in the scrotal skin, dermal skin, kidney, heart, and testis (Figures 3 and 4, Table 1). The inclusions were not seen in the liver, spleen, or muscle. Both neural and nonneural nuclear inclusions were strongly stained by PG-21 and AR(N-20), but not by 2F12, AR52, 5F4, or AR(C-19) (Table 2). The inclusions were spherical structures of 1 to 5 ␮ m in diameter, and those in the nonneural tissues were generally smaller in size than those in the neurons. They were ubiquitinated (Figure 3, Tables 1 and 2). There were no detectable inclusions in cytoplasm of the neural or nonneural tissues. Most com- monly we saw one inclusion in the nucleus of the individ- ual cells, but two or three inclusions per cell were also observed (Figures 2 to 4) in both the neural and nonneural tissues. The frequency of the nuclear inclusions in the spinal motor neurons was 8.39 Ϯ 5.43% of total remain- ing motor neurons. The frequencies of nuclear inclusions in nonneural tissues were 0.93 Ϯ 0.50% in scrotal skin epidermal cells, 0.40% in nonscrotal skin epidermal cells, and 0.59 Ϯ 0.02% in kidney tubular cells, and nuclear inclusions were observed only occasionally in the heart and testis. Electron microscopically, these inclusions consisted of granular dense aggregates of AR-positive materials without limiting membrane (Figure 4), and the morphological appearance was quite similar among the inclusions in the spinal motor neurons, scrotal skin, and kidney. There was no evidence of filamentous structures as reported in HD 27 and MJD 30 (Figure 4). The morphological appearance of the motor neurons and nonneural cells with nuclear inclusions was indistin- guishable from those without inclusions. Neural and nonneural tissues from four control cases were also examined in the same manner as that for SBMA cases, but the inclusions were not seen in the control individuals. The present study demonstrates that nuclear inclusions of the mutant AR protein are present in the motor neurons of the spinal cord and the brain stem; moreover, similar inclusions are also detected in certain nonneural tissues of the SBMA patients, indicating that the occurrence of the nuclear inclusions is not restricted to the affected neural tissues, but is also found in certain nonneural tissues. The electron microscopic appearance of the AR-positive dense aggregates with similar granular size without limiting membrane was common to both neural and nonneural tissues. Furthermore, a characteristic selective staining pattern of the nuclear inclusions seen with only Abs recognizing N-terminal 20 and 21 amino acids of the AR protein was also common among the neural and nonneural tissues. These observations strongly suggest that same mechanism is involved in formation of AR- positive components in the nuclear inclusions in both the neural and nonneural tissues. The immunostaining pattern suggests that only a small portion of the N terminus of the AR protein is available as an epitope in the nuclear inclusions, and other portions of the AR may be masked within the inclusions of the mutant protein, or alternatively, the AR protein may be cleaved by proteolytic activity resulting in N-terminal fragments that participate in the aggregation, as was suggested in other polyglutamine diseases. 26 –36,41 In addition, these AR nuclear inclusions are ubiquitinated in the nonneural as well as neural tissues, indicating that the nuclear inclusion is a pathological structure of the mutant AR, even in the nonneural tissues, where the pathological involvement is not appar- ent. Our electron microscopic and light microscopic immunohistochemical data indicate that nuclear inclusions in both motor neurons and nonneural tissues are identical in morphological and immunochemical features. Furthermore, absence of filamentous structures in the nuclear inclusion of SBMA was different from observations in HD 27 and MJD. 30 This difference may suggest that the pathway of the nuclear aggregation is different among the different protein products or, alternatively, may rep- resent variances in sample preparation. In polyglutamine diseases that have been analyzed to date, the nuclear inclusions have been shown to occur selectively in neurons of the affected brain regions. The selective occurrence of nuclear inclusions in the affected cells of central nervous system in SBMA that we observed in this study agrees well with observations of HD, MJD, SCA1, and DRPLA, 27–32 as well as our previous observations in SBMA. However, the appearance of similar nuclear inclusions in the nonneural tissues observed in this study is novel. It is an important question why the neurons are selectively affected despite the presence of nuclear inclusions in both the affected neurons and nonaffected nonneural tissues. The cells of the nonneural tissues are mitotic cells in contrast to motor neurons; the epidermal cells in the scrotal and dermal skin and epithelial cells in the kidney tubules are all capable of mitosis in adulthood, and are eventually replaced by newly generated cells. Hence, those cells with toxic effects associated with the nuclear inclusions may be replaced by turnover. It may also be that neurons are particularly susceptible to whatever deleterious effects the inclusions may have. As demonstrated in the HD transgenic model, 34 a long latent period is necessary for inclusions to induce neuronal death. Neurons, as postmi- totic cells, may be specifically affected because they survive long enough for the inclusions to have effect, whereas nonneural cells with nuclear inclusions turn over before the inclusions have pathological consequences. The significantly lower frequency of nuclear inclusion in nonneural tissues than in neurons may support this view. These differences in cell turnover rates could contribute to selective neuronal degeneration and neuronal loss. Another interesting observation in this study is that the presence of nuclear inclusions is also selective among the various nonneural tissues, as it is in neural tissues. The inclusions are frequent in scrotal skin, dermal skin, and kidney; only occasionally seen in the testis and heart muscle; and not detected in spleen, liver, and muscle. The distribution of inclusions is not related to the expression level of mutant AR in these ...

Context 3

... a re- combinant protein of 321 amino acids from the N terminus of the human AR; PG-21 (rabbit polyclonal Ab (IgG), Affinity BioReagents, Golden, CO) and AR(N-20)(rabbit polyclonal Ab (IgG), Santa Cruz Biotechnology, Santa Cruz, CA), which recognize 21 and 20 amino acid resi- dues of the N terminus of the AR, respectively; AR52 (rabbit polyclonal Ab (IgG), kindly provided by Dr. E. Wilson, University of North Carolina, Chapel Hill, NC), which recognizes the DNA-binding domain of the AR; and 5F4 (mouse monoclonal Ab (IgM), kindly provided by Dr. T Demura, Department of Urology, Hokkaido Univer- sity, Hokkaido, Japan) and AR(C-19) (rabbit polyclonal Ab (IgG), Santa Cruz Biotechnology), which recognizes the C terminus of the human AR. Characterization and binding specificities of all of these Abs to human AR were previously described. 33,37–39 Anti-ubiquitin Ab (rabbit polyclonal Ab (IgG), Dakopatts, Glostrup, Denmark) was also used. Cryostat sections of 8 ␮ m were prepared from the frozen tissues of SBMA patients and controls, quickly dried, and lightly fixed with Zamboni fixative for 10 minutes. Then the tissue sections were washed, blocked with normal horse serum (1:20), and incubated with Abs against AR, ubiquitin, or affinity-purified mouse IgG1 at concentrations of 0.5 to 4 ␮ g/ml. Endogenous peroxidase was blocked by preincubation of tissue sections with 0.3% H 2 O 2 in meth- anol for 30 minutes. Endogenous biotin was also blocked by incubation with an avidin-biotin blocking kit (Vector Laboratories, Burlingame, CA). Immune complexes were visualized using the avidin-biotinylated horseradish peroxidase system (Elite Vector kit from Vector Laboratories) and 3,3 Ј -diaminobenzidine (Dakopatts) substrate. Sec- tions were counterstained with methyl green. For paraffin- embedded samples, the 4- ␮ m tissue sections were deparaffinized in xylene, hydrated with alcohol, and then heated in a microwave oven for 5 minutes. The tissue sections were then processed in the same way as those for the frozen tissue sections. Immune complexes were visualized using the tyramide signal amplification system (New England Nuclear, Boston, MA) following the manu- facturer’s protocol. For electron microscopic immunohistochemistry, buffered formalin-fixed, paraffin-embedded tissue sections were immunostained with Abs against AR and then incubated with horseradish peroxidase-labeled second Ab (Amersham, Poole, UK). The tissue sections were then visualized with 3,3 Ј -diaminobenzidine (Dakopatts), fixed with 2% osmium tetroxide in 0.1 mol/L phosphate buffer, pH 7.4, and dehydrated in an alcohol gradient and embedded in epoxy resin, from which ultrathin sections were obtained and then observed under an electron micro- scope (Hitachi H-7000). CAG repeat length of the AR gene was determined on the autopsied tissue samples using methods described previously. 5,40 Nuclear inclusions were detected in the spinal motor neurons, neurons in the motor nucleus of the trigeminal nuclei in the pons, and neurons of the hypoglossal nuclei of the medulla oblongata (Figure 2 (see also Figure 4), Table 1). Nuclear inclusions were not detected in the neurons of cerebral cortex, caudate, thalamus, pallidum, cerebellar cortex, dentate nucleus nuclei other than motor nuclei in the brain stem, intermediolateral and Clarke’s nuclei of the spinal cord, or dorsal root ganglion neurons. Nuclear inclusions similar to those in the motor neurons were detected in the scrotal skin, dermal skin, kidney, heart, and testis (Figures 3 and 4, Table 1). The inclusions were not seen in the liver, spleen, or muscle. Both neural and nonneural nuclear inclusions were strongly stained by PG-21 and AR(N-20), but not by 2F12, AR52, 5F4, or AR(C-19) (Table 2). The inclusions were spherical structures of 1 to 5 ␮ m in diameter, and those in the nonneural tissues were generally smaller in size than those in the neurons. They were ubiquitinated (Figure 3, Tables 1 and 2). There were no detectable inclusions in cytoplasm of the neural or nonneural tissues. Most com- monly we saw one inclusion in the nucleus of the individ- ual cells, but two or three inclusions per cell were also observed (Figures 2 to 4) in both the neural and nonneural tissues. The frequency of the nuclear inclusions in the spinal motor neurons was 8.39 Ϯ 5.43% of total remain- ing motor neurons. The frequencies of nuclear inclusions in nonneural tissues were 0.93 Ϯ 0.50% in scrotal skin epidermal cells, 0.40% in nonscrotal skin epidermal cells, and 0.59 Ϯ 0.02% in kidney tubular cells, and nuclear inclusions were observed only occasionally in the heart and testis. Electron microscopically, these inclusions consisted of granular dense aggregates of AR-positive materials without limiting membrane (Figure 4), and the morphological appearance was quite similar among the inclusions in the spinal motor neurons, scrotal skin, and kidney. There was no evidence of filamentous structures as reported in HD 27 and MJD 30 (Figure 4). The morphological appearance of the motor neurons and nonneural cells with nuclear inclusions was indistin- guishable from those without inclusions. Neural and nonneural tissues from four control cases were also examined in the same manner as that for SBMA cases, but the inclusions were not seen in the control individuals. The present study demonstrates that nuclear inclusions of the mutant AR protein are present in the motor neurons of the spinal cord and the brain stem; moreover, similar inclusions are also detected in certain nonneural tissues of the SBMA patients, indicating that the occurrence of the nuclear inclusions is not restricted to the affected neural tissues, but is also found in certain nonneural tissues. The electron microscopic appearance of the AR-positive dense aggregates with similar granular size without limiting membrane was common to both neural and nonneural tissues. Furthermore, a characteristic selective staining pattern of the nuclear inclusions seen with only Abs recognizing N-terminal 20 and 21 amino acids of the AR protein was also common among the neural and nonneural tissues. These observations strongly suggest that same mechanism is involved in formation of AR- positive components in the nuclear inclusions in both the neural and nonneural tissues. The immunostaining pattern suggests that only a small portion of the N terminus of the AR protein is available as an epitope in the nuclear inclusions, and other portions of the AR may be masked within the inclusions of the mutant protein, or alternatively, the AR protein may be cleaved by proteolytic activity resulting in N-terminal fragments that participate in the aggregation, as was suggested in other polyglutamine diseases. 26 –36,41 In addition, these AR nuclear inclusions are ubiquitinated in the nonneural as well as neural tissues, indicating that the nuclear inclusion is a pathological structure of the mutant AR, even in the nonneural tissues, where the pathological involvement is not appar- ent. Our electron microscopic and light microscopic immunohistochemical data indicate that nuclear inclusions in both motor neurons and nonneural tissues are identical in morphological and immunochemical features. Furthermore, absence of filamentous structures in the nuclear inclusion of SBMA was different from observations in HD 27 and MJD. 30 This difference may suggest that the pathway of the nuclear aggregation is different among the different protein products or, alternatively, may rep- resent variances in sample preparation. In polyglutamine diseases that have been analyzed to date, the nuclear inclusions have been shown to occur selectively in neurons of the affected brain regions. The selective occurrence of nuclear inclusions in the affected cells of central nervous system in SBMA that we observed in this study agrees well with observations of HD, MJD, SCA1, and DRPLA, 27–32 as well as our previous observations in SBMA. However, the appearance of similar nuclear inclusions in the nonneural tissues observed in this study is novel. It is an important question why the neurons are selectively affected despite the presence of nuclear inclusions in both the affected neurons and nonaffected nonneural tissues. The cells of the nonneural tissues are mitotic cells in contrast to motor neurons; the epidermal cells in the scrotal and dermal skin and epithelial cells in the kidney tubules are all capable of mitosis in adulthood, and are eventually replaced by newly generated cells. Hence, those cells with toxic effects associated with the nuclear inclusions may be replaced by turnover. It may also be that neurons are particularly susceptible to whatever deleterious effects the inclusions may have. As demonstrated in the HD transgenic model, 34 a long latent period is necessary for inclusions to induce neuronal death. Neurons, as postmi- totic cells, may be specifically affected because they survive long enough for the inclusions to have effect, whereas nonneural cells with nuclear inclusions turn over before the inclusions have pathological consequences. The significantly lower frequency of nuclear inclusion in nonneural tissues than in neurons may support this view. These differences in cell turnover rates could contribute to selective neuronal degeneration and neuronal loss. Another interesting observation in this study is that the presence of nuclear inclusions is also selective among the various nonneural tissues, as it is in neural tissues. The inclusions are frequent in scrotal skin, dermal skin, and kidney; only occasionally seen in the testis and heart muscle; and not detected in spleen, liver, and muscle. The distribution of inclusions is not related to the expression level of mutant AR in these tissues. 33 AR protein is highly expressed in the testis, skin, and muscle, whereas it is low in the kidney, spleen, and liver. 33 A similar lack of correlation between the AR protein expression levels and pathological ...